Deep learning based vehicle license plate position insensitive vehicle license plate recognition method

Abstract

The invention discloses a deep learning based vehicle license plate position insensitive vehicle license plate recognition method, which is characterized in that a sample set with seven vehicle license plate character labels is built, and training is carried out on a deep convolution neural network. The vehicle license plate recognition method comprises the steps of carrying out preprocessing on a vehicle license plate image to be detected, converting the preprocessed vehicle license plate image into an image whose size is the same as that of a sample in the sample set, and inputting the converted image into the trained deep convolution neutral network; carrying out one time of forward propagation in the deep convolution neural network, and outputting seven labels; and acquiring Chinese characters and characters corresponding to the labels through looking up a table so as to acquire seven vehicle license plate characters. According to the invention, vehicle license plate position information can be effectively recognized through creatively sharing a convolution layer, carrying out recognition on the seven vehicle license plate characters and carrying out sample processing in a targeted manner at the same time, thereby improving the recognition rate under a complex environment, and finally achieving a good global recognition effect.

Description

technical field [0001] The invention relates to the technical fields of computer vision and machine learning, in particular to a deep learning-based license plate recognition method that is insensitive to the position of the license plate. Background technique [0002] Intelligent traffic monitoring system is a key development direction of today's traffic monitoring industry. It mainly relies on computer vision and other technologies to automatically analyze the pictures captured by surveillance cameras, so as to judge illegal behaviors such as speeding and running red lights, and can automatically identify them. The license plate number of illegal vehicles, etc., which greatly facilitates traffic supervision. [0003] Deep learning was officially proposed in 2006. It is a popular field in machine learning in recent years. It originated from multi-layer artificial neural networks and has been successfully applied in computer vision, natural language processing and intelligen...

AI Technical Summary

Problems solved by technology

[0004] In the field of license plate recognition, the traditional method usually uses an image processing algorithm to locate the license plate, and then extracts a single character through image segmentation, and then uses a recognition method based on character strokes for character recognition. This method can also achieve good results, but It is cumbersome to use, and has high requirements for preprocessing and segmentation steps. If the image quality is poor or the segmentation effect is not good, the recognition effect cannot be guaranteed.

[0005] After searching the existing technologies, some methods of license plate recognition using convolutional neural networks have also been found. However, in the processing of multiple characters, it is often necessary to rely on traditional image segmentation algorithms to segment a single character and then They are sent to the convolutional neural network for recognition separately, so that the requirement for the segmentation effect cannot be avoided, and only limited effect improvement can be obtained.

On the other hand, the existing license plate recognition methods usually first determine the position of the license plate through the detection algorithm, and then only input the image of the license plate part to the recognition algorithm for license plate recognition, which puts forward higher requirements for the previous detection algorithm. Require

Images